A NOx Prediction Scheme for Lean-Premixed Gas Turbine Combustion Based on Detailed Chemical Kinetics

Abstract

The lean-premixed technique has proven very efficient in reducing the emissions of oxides of nitrogen (NOx) from gas turbine combustors. The numerical prediction of NOx levels in such combustors with multidimensional CFD codes has only met with limited success so far. This is to some extent due to the complexity of the NOx formation chemistry in lean-premixed combustion, i.e., all three known NOx formation routes (Zeldovich, nitrous, and prompt) can contribute significantly. Furthermore, NOx formation occurs almost exclusively in the flame zone, where radical concentrations significantly above equilibrium values are observed. A relatively large chemical mechanism is therefore required to predict radical concentrations and NOx formation rates under such conditions. These difficulties have prompted the development of a NOx postprocessing scheme, where rate and concentration information necessary to predict NOx formation is taken from one-dimensional combustion models with detailed chemistry and provided—via look-up tables—to the multidimensional CFD code. The look-up tables are prepared beforehand in accordance with the operating conditions and are based on CO concentrations, which are indicative of free radical chemistry. Once the reacting flow field has been computed with the main CFD code, the chemical source terms of the NO transport equation, i.e., local NO formation rates, are determined from the reacting flow field and the tabulated chemical data. Then the main code is turned on again to compute the NO concentration field. This NOx submodel has no adjustable parameters and converges very quickly. Good agreement with experiment has been observed and interesting conclusions concerning superequilibrium O-atom concentrations and fluctuations of temperature could be drawn.